On Wed, Jul 17, 2019 at 11:41 AM James(王旭) <wan...@gu360.com> wrote:
> From these results I can tell the route to a table is not even related with
> the mod function, right?
> So It's hard for me to do any kind of guesses...
Because it is the wrong function.
According to \d+ on a child table and partbounds.c the function called
is satisfied_hash_partition:
testdb=# select satisfies_hash_partition('153221'::oid, 3, 0, 6521);
satisfies_hash_partition
--------------------------
t
(1 row)
testdb=# select satisfies_hash_partition('153221'::oid, 3, 1, 6521);
satisfies_hash_partition
--------------------------
f
(1 row)
The first argument is the table id (partitioned one, the root), the
second is the reminder, third is the partition table, last is your
value.
Therefore I suspect you have to iterate on your partition numbers from
0 to x to see if a value fits in that partition, and then extract the
table name from that.
Hope its clear.
Luca
